Free AI Stem Splitter

Free stem splitter online. A stem splitter separates a finished song into its parts so you can remix, sample, practise, or make a karaoke track. This free online stem splitter uses a real AI source separation model to produce four stems from any track: vocals, drums, bass, and other instruments. It also builds a ready instrumental with the vocals removed. Preview each one and download it as a WAV.

It runs in your browser, so your track is processed on your device and never uploaded.

The tool runs Demucs, an open AI music separation model from Meta research, directly in your browser. The neural network was trained to recognise how voices, drums, bass, and other instruments sound, so it can pull them apart even when they overlap in the mix. This is the same family of technology behind well known stem separation services.

The first time you use it, the model downloads once and is saved by your browser for next time. Splitting is fastest in a browser with GPU acceleration, such as a recent Chrome or Edge on a computer. Mono and stereo files both work.

If you want stems you fully own and control, generating a song from scratch beats splitting an existing one. InstaSong creates original tracks, with or without vocals, in the style you choose, so you avoid copyright issues and bleed between parts.

It is free to start with credits included, which is a fast way to get clean, original material for your project.